1 Revision history for the Perl extension Time::HiRes.
4 - This version reverts the changes on CPAN that were included in 1.9754 - 1.9758.
5 Several of these changes caused instability on some windows platforms. We can
6 revisit these fixes if necessary in Perl 5.29.0
9 - in t/clock.t in a fast system we need to burn more CPU,
10 reported and fix suggested by Joel C. Maslak <jmaslak@antelope.net>
12 - for t/utime.t in netbsd we need /sbin in PATH to find the mount command,
13 this is needed for the fix in 1.9751 to actually help netbsd with noatime
15 (thanks to Nigel Horne)
18 - fix an error in the error message of utimensat() not available:
19 it said futimens() not available
20 - add --force alias for Makefile.PL --configure
23 - in macOS/OSX/Darwin, use __has_builtin() check also for utimensat(),
25 "HiRes.xs:1474:16: error: unrecognized platform name macOS"
27 (oversight from 1.9749)
28 - do not define TIME_HIRES_STAT/d_hires_stat if none was found, instead
29 of defining it to be zero, which case has no implementation in hrstatns()
30 (thanks to Nigel Horne)
31 - in t/utime.t try to divine if the filesystem of the tempfiles has been
32 mounted with the 'noatime' option, which can prohibit updating the
33 access time timestamp. Also document this in HiRes.pm.
34 (thanks to Nigel Horne, original analysis by Slaven Rezic)
35 - synchronize the constant lists in HiRes.pm:@EXPORT_OK
36 and Makefile.PL:doConstants and regenerate fallback/const-c.inc
37 and fallback/const-xs.inc, this fixes Perl 5.6.2 issue with
38 d_futimens not allegedly being a valid macro in t/utime.t
39 (using Perl 5.26.1 for the regenerating, not 5.6.2)
40 (thanks to Nigel Horne)
41 - in t/utime.t define a nop sub done_testing for ancient Perls
43 - in Perl 5.6.2 a bogus warning
44 "Use of uninitialized value in subroutine entry"
45 is issued from t/alarm.t: add a comment documenting that
48 - update META.yml and META.json for XSLoader instead of DynaLoader
53 - Quieten Time-HiRes test noise: blead 0f009486
54 - Switch some core modules to XSLoader: blead b9a5a78f [rt.cpan.org #129789]
55 - disable a flapping test for CI: blead cd2a978d
56 - Avoid newGVgen in blead-upstream modules: blead 732d3893
57 - Declaration after statement in typemap: blead d269f586
58 - semicolon-friendly diagnostic control: blead 7347ee54
59 - in macOS (OS X) check properly in runtime whether futimens()
60 and utimensat() are available
64 - semicolon-friendly diagnostic control: blead 7347ee54
68 - Switch some core modules to XSLoader: blead b9a5a78f [rt.cpan.org #129789]
71 - Unreliable t/usleep.t and t/utime.t tests [rt.cpan.org #122819]
72 Avoid testing for $dt = $t2 - $t1 and assuming $dt is less than
73 some value since a heavily loaded machine can delay the $t2.
74 Testing for greater than is fine.
75 - Tweak the configuring messages about subsecond stat/utime.
78 - Properly check for futimens/utimensat actually doing something.
79 This should handle 'gnukfreebsd' (which has only stubs, so far
80 we have been skipping the test) and as a new thing 'gnu' (Hurd)
81 (also only stubs). Thanks to Nigel Horne.
82 - Scan in t/utime.t whether the filesystem (wherever File::Temp
83 puts it tempfiles) supports subsecond timestamps. This removes
84 the fragile Linux/ext3 specific hack. As a side effect, the
85 setting of subsecond timestamps is tested only if reading of
86 them is supported. Thanks to Carsten Gaebler for the test idea,
87 and Ryan Voots for testing.
90 - add more potential clock constants, like CLOCK_MONOTONIC_FAST
91 (available in FreeBSD), and not all potentially found clock
92 constants were properly exported to be available from Perl,
93 see your system's clock_gettime() documentation for the available ones
96 - correct declared minimum Perl version (should be 5.6, was declared
97 as 5.8 since 1.9727_03): blead af94b3ac
98 - fix the fix for 'do file' to load hints in Makefile.PL: blead 3172fdbc
101 - prefer 3-argument open: blead 1ae6ead9
102 - fix dist/Time-HiRes/t/*.t that assumed '.' in @INC: blead 465db51d
103 - fix cases where 'do file' should be 'do ./file'.: blead 8b69401c
106 - C++11 compatibility: blead a914236c
107 - El Capitan compatibility: blead 45bbc013
108 - use CLOCK_REALTIME for clock_nanosleep scan: blead 56ed632c
109 - include file consistency in scans: blead 605b4dbe
110 - use clockid_t consistently: blead 2979fdf3
111 - use hv_fetchs(): blead 07e4dd7a
112 - scan for clockid_t (needed for macos Sierra): blead 69d1f2c2
113 - darwin lacks clockid_t: blead d1e7b5aa
114 [rt.cpan.org #129789]
115 - provide template for tempfile: blead 91dca83e
116 - explicit cast for clockid_t in C++11: blead 1c566547
117 - upgrade to a newer ppport (3.35)
120 - the ext3/ext2 filesystems do not have subsecond resolution,
121 therefore skip the t/utime.t test
122 [rt.cpan.org #116127]
125 - the upcoming macOS 10.12 (Sierra, the operating system formerly
126 known as OS X, or Darwin) has implemented the clock_gettime()
127 and clock_getres() interfaces but not the clock_nanosleep()
130 - fix the dates in the Changes, it's not July yet
131 [rt.cpan.org #115553]
132 - in cygwin, drop the utime.t expected timestamp resolution
133 - documented for utime that the actual timestamp resolution depends
134 on the operating system and the filesystem, just like for stat
135 - looks like in gnukfreebsd 10.1 neither futimens or utimensat work
136 even though they exist, skipping utime tests there for now
137 - demodernize the Test::More use in utime.t (no subtest) to
138 better support older Perl installations
139 - for all the tests, drop the Test::More 0.82 requirement,
140 the only "modern" thing used is the note routine which
141 can be replaced with print("# ...\n") (or printf),
142 with this works back at least with Test::More 0.54 (perl 5.8.7)
145 - utime in HiRes.xs had dXSARGS which was unnecessary, leading
146 to two markstack pops, panic, and two-minute wait for the watchdog
147 (thanks to Dave Mitchell for the eagle-eye)
148 - even though 5.10.x should have the Perl_ck_warner, many CPAN
149 smokes proved otherwise, so requiring now at least 5.12 for
150 Perl_ck_warner use in IS_SAFE_PATHNAME()
153 - the IS_SAFE_PATHNAME() emulation was too modern for Perl 5.10
154 - make it so that only one value is set for -DTIME_HIRES_STAT
155 even on systems that support many options
156 - the 1.9734 (only released in blead, and as part of the development
157 release 5.25.2) fix for Darwin's clock_gettime was wrong under
158 threaded builds, leading into mutex panics in clock.t
159 - the d_hires_utime determination was made more precise
160 - the utime.t was failing in platforms where only one of
161 futimens/utimensat was present (neither or both was okay)
164 - Time::HiRes should override `utime` to allow setting hires
165 (futimens and utimensat support)
166 [rt.perl.org #114809]
167 - the utime patch used IS_SAFE_PATHNAME() which isn't available in
168 too old Perls, so emulate (in case the Devel::PPPort is too old)
171 - fix Darwins with clock_gettime: blead 2d41a263
172 [rt.perl.org #128427]
175 - C90 declaration-after-statement error with darwin threads: blead de1003b4
176 [rt.cpan.org #113856]
179 - MUTEX_LOCK, not PERL_MUTEX_LOCK: blead e5b02b5d
180 - also hrt_ualarm_itimer() is unused: 1cb6cce3
181 - the mutex needs init: 2d639e20
182 - version bump to 1.9732: 730d7fdc
185 - mark unused variable as such: blead a914236c
186 - OS X emulation mutex accidentally unused: da7a6455
187 - remove hrt_ualarm: 6da77c36
188 - pod error fixes: 919ca095
189 - nanosleep and clock_nanosleep cleanups: c8ea02b8..e3ff671b
190 - static funcs in HiRes.xs: 4e922c26
191 - Remove unused variable: 52ffb1b5
194 - TIME_HIRES_DONT_RUN_PROBES=1 to build the probes but not run them
195 [rt.cpan.org #111391]
199 [rt.cpan.org #111170]
200 - clock_getres(), clock_gettime(), clock_nanosleep() emulation for OS X
204 - No changes since 1.9727_03
206 1.9727_03 [2015-11-06]
207 - use PERL_UNUSED_ARG() in the not-here-croak branches
208 - Makefile.PL/metadata improvements
210 1.9727_02 [2015-08-29]
211 - correct documentation to note that Time::NVtime is NV, not double
213 - avoid a warning about clock_t on OS X / clang 3.6 - 3.6
214 [rt.cpan.org #102718]
216 1.9727_01 [2015-08-17]
217 - Remove usage of 5.005 thread APIs in favor of 5.6 and up thread APIs
218 [rt.cpan.org #102799].
219 - Silence a C++ compiler warning [rt.cpan.org #101497].
220 - respect $Config{run} for cross-compiling [rt.cpan.org #92607]
223 - Correct s/us splitting of usleep(1000000) [rt.cpan.org #78266].
224 - Avoid integer overflow in itimer-based alarm() with large
225 argument [rt.cpan.org #87160].
226 - Define PERL_NO_INLINE_FUNCTIONS during probe compilations, to
227 avoid false negatives caused by not linking with the perl core.
228 - Be more careful about context in stat().
229 - Install into "site" directories by default on perl 5.11+
230 [rt.cpan.org #79797].
231 - Fix a couple of doc typos [rt.cpan.org #85365].
232 - Fix function name in a doc example [rt.cpan.org #86318].
233 - Provide lstat() that yields high-res timestamps, alongside
234 the existing high-res stat() [rt.cpan.org #78732].
237 - Correct stack discipline in stat(), which was screwing up list
238 operations in expressions containing calls to it [rt.cpan.org
240 - Add missing OUTPUT sections to the XS code [rt.cpan.org #70930].
241 - Skip itimer tests on GNU/Hurd, which has the API but lacks
242 the implementation [rt.cpan.org #72754].
243 - Fix a doubled word in the documentation [rt.cpan.org #72763].
246 - Correct XS parameter list, and therefore prototype, for
247 unimplemented-on-this-platform version of clock_nanosleep()
248 [rt.cpan.org #68700].
249 - Declare package variables with "our" rather than "use vars".
250 - Corresponding to "our" usage, check for minimum Perl version
252 - Declare module dependencies.
255 - Remove $ENV{PERL_CORE} logic from test suite, which is no
256 longer desired in the core.
257 - Convert test suite to use Test::More.
258 - Factor out watchdog code from test suite.
259 - In test suite, be consistent about using fully-qualified form
261 - Divide test suite into feature-specific scripts.
262 - Make ualarm timing test less vulnerable to delay-induced false
263 failure, from Dave Mitchell.
266 - Update for changes in build process in the core, patches
267 from BinGOs [rt.cpan.org #58858] and Craig Berry [rt.cpan.org
269 - Fix broken linkage on Windows with gcc 3.4 seen with ActivePerl,
270 report from Christian Walde [rt.cpan.org #61648], fix derived
272 - Jump through hoops to avoid compiler warnings.
275 - Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
276 additional fix from Gisle Aas
277 - Address [rt.cpan.org #55665] "Bad plan" on Windows,
278 report and fix from Jan Dubois
281 - Address [rt.cpan.org #54196] alarm and ularm return values are bogus,
282 report and fix from Nicholas Clark
285 - As with QNX, Haiku has the API of interval timers but not
286 the implementation (bleadperl change #34630), hence skip
287 the tests, via David Mitchell.
290 - .xs code cleanup from Albert Dvornik
291 - in the #39 and #40 do not do us I did, mixing alarm() and
292 sleep(). Now instead spin until enough time has passed.
295 - Skip the tests added in 1.9716 (#39, #40) if there's no subsecond
296 alarm capability, like with the older subsecond alarm tests
299 - Change documentation to agree with reality: there are
300 no interval timers in Win32.
301 - Address [rt.cpan.org #35899] (problem in subsecond sleeps),
302 add two tests to guard against this problem
303 - Address [rt.cpan.org #36600] 'Division by zero' failure in test suite
304 - Address [rt.cpan.org #37340] [PATCH] Address timer process in test
305 - Address [rt.cpan.org#40311 ] bad implementation of hrt_usleep
306 with TIME_HIRES_NANOSLEEP
309 - Silly me: Makefile.PL does need to accept arguments other than mine.
310 Some testing frameworks obviously do this.
311 - Add retrying for tests 34..37, which are the most commonly
312 failing tests. If this helps, consider extending the retry
313 framework to all the tests. [Inspired by Slaven Rezic,
314 [rt.cpan.org #34711] Occasional failures of test 35 or 36 (FreeBSD)]
317 - Under Perl 5.6.* NVgf needs to be "g", reported by Zefram,
318 it seems that ppport.h 3.13 gets this wrong.
319 - remove the check in Makefile.PL for 5.7.2, shouldn't be
320 (a) necessary (b) relevant
321 - add logic to Makefile.PL to skip configure/write Makefile
322 step if the "xdefine" file already exists, indicating that
323 the configure step has already been done, one can still
324 force (re)configure by "perl Makefile.PL configure",
325 or of course by "make clean && perl Makefile.PL".
328 - for alarm() and ualarm() [Perl] prefer setitimer() [C]
329 instead of ualarm() [C] since ualarm() [C] cannot portably
330 (and standards-compliantly) be used for more than 999_999
331 microseconds (rt.cpan.org #34655)
332 - it seems that HP-UX has started (at least in 11.31 ia64)
333 #defining the CLOCK_REALTIME et alia (instead of having
335 - document all the diagnostics
338 - move the sub tick in the test file back to where it used to be
339 - in the "consider upgrading" message recommend at least Perl 5.8.8
340 and make the message to appear only for 5.8.0 since 5.8.1 and
341 later have the problem fixed
342 - VOS tweak for Makefile (core perl change #33259)
343 - since the test #17 seems to fail often, relax its limits a bit
346 - lost VMS test skippage from Craig Berry
347 - reformat the test code a little
350 - I got the sense of the QNX test the wrong way in an attempt
351 to generalize it for future
354 - casting fixes from Robin Barker for g++ and 64bitint
355 - in QNX skip the itimer tests because though the API
356 is there, the implementation isn't, from Matt Kraai
357 - raise the dead man timer to 180 seconds for really
359 - elaborate the UTF-8 locale warning from Makefile.PL
362 - [rt.cpan.org #29747]: Build failure with perl 5.005_05
363 Fixed by regenerating the ppport.h using Devel::PPPort 3.13.
366 - t/HiRes.t failed in Perl 5.6.2,
367 "action is not of type POSIX::SigAction at t/HiRes.t line 318",
368 reported and fixed by Anton Berezin, the reason was faulty
369 use of sigaction() when restoring the old action.
372 - with bleadperl in VMS the HiRes.t overrun the maximum number
373 of deferred signals because the libc SIGALRM was not strong
374 enough to interrupt select(), and select() got restarted every
375 time, solution is to use POSIX::SigAction if available.
376 A fix from Craig Berry (not 100% there, but helps).
377 - allow for more measuring noise for ualarm() tests 35..37
380 - nanosleep() and clock_nanosleep() detection and use were
381 quite broken; in Linux -lrt needed; fixes from Zefram
382 - [internal] slightly cleaner building of $DEFINE in Makefile.PL,
383 should avoid double/conflicting -D flags
386 - allow 10% of slop in test #14 (testing difference between
387 CORE::time() and Time::HiRes::time()), there seem to be often
388 transient failures from Perl smoke builds on this test
392 - use int main(int argc, char **argv) consistently in Makefile.PL,
394 [rt.cpan.org #23868] nanosleep not detected under Mac OS 10.3.9 starting with Time::HiRes 1.96
395 - if someone still has the locale-broken Perl 5.8.0,
396 suggest that they upgrade their Perl
399 - restore the -DATLEASTFIVEOHOHFIVE, Win32 needed it still
402 - upgrade to ppport.h 3.10_02
403 - remove the -DATLEASTFIVEOHOHFIVE
404 - use the ppport.h PL_ppaddr, PL_statcache, PL_laststatval
405 - use the ppport.h aTHXR for calling Perl stat()
406 - switch into four-digit version since 2.0 is coming up
407 awfully fast but not feeling like a major rewrite
410 - 1.95 broke building in Win32 (since pp_stat is not exported),
411 figured out how to call an op directly in 5.005 (use Perl_ppaddr
412 instead of PL_ppaddr)
413 - backport to Perl 5.004_05 (requires using statcache
414 and laststatval instead of PL_statcache and PL_laststatval)
415 (also checked to work in 5.005_04, 5.6.1, and 5.8.8 with threads)
418 - 1.95 broke builds for threaded Perls, rt.cpan.org tickets:
419 [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
420 [rt.cpan.org #23712] Time-HiRes 1.95 Fails make on AIX 5.2 with Perl 5.8.8
421 [rt.cpan.org #23730] Time::HiRes 1.95 fails make on MacOS X 10.3.9/perl 5.8.8
422 - use main() prototype consistently in Makefile.PL
425 - integrate core change #29180: Silence VC++ compiler warnings
427 - do not use PL_ppaddr in stat() because that is not available
429 - regenerate fallback/*.inc for older Perls without
430 ExtUtils::Constant because of d_hires_stat, resolves
431 [rt.cpan.org #23694] Time::HiRes fails tests on Solaris and Perl 5.6.1
432 - Make Makefile.PL more defensive against false PERL_CORE
435 - file timestamps oddities seen: the atime and mtime
436 can be out of sync (modify first and read second can leave
437 atime < mtime) and mtime can be subsecond while atime is not.
438 So make the test more forgiving.
441 - the ualarm() tests (34-37) assumed that ualarm(N)
442 could never alarm in less than N seconds, widened
443 the acceptable relative range to 0.9..1.5. Addresses
444 [rt.cpan.org #22090] and [rt.cpan.org #22091].
446 - skip the stat() tests in cygwin and win32, because
447 if run on FAT the timestamp granularity is only 2 seconds.
448 Any good way to detect (cygwin or win32) whether we are
449 being run on NTFS or anywhere with better timestamps?
450 Addresses [rt.cpan.org #22089] and [rt.cpan.org #22098].
453 - scan for subsecond resolution timestamps in struct stat,
454 some known possibilities:
456 (1) struct timespec st_atimespec;
457 st_atimespec.tv_nsec;
462 (4) timestruc_t st_atim;
467 If something like this is found, one can do
470 my @stat = Time::HiRes::stat();
472 or even override the standard stat():
474 use Time::HiRes qw(stat);
476 to get the stat() timestamps
478 my ($atime, $mtime, $ctime) = @stat[8, 9, 10];
480 with subsecond resolution (assuming both the operating
481 system and the filesystem support that kind of thing).
483 Contributions for more systems (especially non-UNIX,
484 e.g. but not limited to: Win32, VMS, OS/2) gladly accepted.
485 (also more UNIX variants welcome: HP-UX? IRIX?)
487 Thanks to H.Merijn Brand, John Peacock, and Craig
488 Berry for brave beta testing.
491 - ualarm() in SuSE 10.1 was overflowing after ~4.2 seconds,
492 possibly due to a glibc bug/feature (suspected overflow at
493 2**32 microseconds?), workaround by using the setitimer()
494 implementation of ualarm() if either useconds or
495 interval > 999_999 (this case seems to vary between systems:
496 are useconds more than 999_999 for ualarm() defined or not)
497 Added more ualarm() tests to catch various overflow points,
498 hopefully no problems in various platforms.
499 (The problem report by Mark Seger and Jon Paul Sullivan of HP.)
502 - tweak still needed for Const64(), from Jerry Hedden
503 - get a freshly generated ppport.h
504 - update Copyright years
507 - Const64() already appends an 'LL' (or i64), so provide LL and i64
508 forms for the IV_1E[679] (effects Win32 and Cygwin), reported by
510 - the Changes entry for 1.88 talked about [IN]V_1[679],
514 - clean up the g++ warnings in HiRes.xs, all of them
515 about mixing integer and floating point, introduce
516 constants IV_1E[679] and NV_1E[679]
519 - [rt.cpan.org #17442] 'make test' frequently fails under
520 Cygwin Perl v5.8.8, reported and patched by J. R. Hedden
521 (two race condition bugs in the END block in the case the
522 main process dies before the timer process, unearthed
523 by a bug in Cygwin ualarm)
526 - HiRes.t:s/ok 32/ok 33/, from Dominic Dunlop
527 - tighten up the clock() test marginally by requiring non-negative
528 - clock_nanosleep() and clock() doc tweaks
531 - the interface to clock_nanosleep() is more natural
532 when it is like (hires) time() (instead of like nanosleep),
533 and the .xs implementation of clock_nanosleep() in 1.84
535 - the semantics of clock() are not quite so silly as I thought,
536 but still somewhat odd, documented as such
537 - additional enhancements to the clock() documentation
538 - add test for clock_nanosleep() (I cannot test this
539 since none of my systems have the function)
540 - add test for clock()
543 - add clock() which returns the processor time in
544 (floating point) seconds since an arbitrary era
545 - add clock_nanosleep() which suspends the current
546 thread until either absolute time or for relative time
547 - [rt.cpan.org #16486] printf missing value in HiRes.t
548 - add constants CLOCKS_PER_SEC, CLOCK_SOFTTIME, TIMER_ABSTIME
552 - has_symbol() was wrong since e.g. ITIMER_VIRTUAL is exported
553 via @EXPORT_OK even when it is not available. This is heinous.
554 @EXPORT_OK should be determined at Makefile.PL time.
555 - be more lenient is testing clock_gettime(): allow more slop,
556 and retry up to three times, sleeping a random nap between
558 - human months are one-based (noticed by Anton Berezin)
561 - CLOCK_REALTIME is an enum value (of the clockid_t enum)
562 in HP-UX (and might be so elsewhere, too), debugged by
564 - include const-c.inc as late as possible (from Randy Kobes,
565 [rt.cpan.org #15552] to avoid undefined usleep() on Win32
568 - try to be more robust and consistent in the detection of
569 CLOCK_REALTIME and ITIMER_VIRTUAL in HiRes.t: the proper
574 eval 'import Time::HiRes qw($symbol)';
575 return 0 unless $@ eq '';
576 return exists ${"Time::HiRes::$symbol"};
583 in the test. All these moves are needed because
585 1) one cannot directly do eval 'Time::HiRes::FOO_BAR'
586 because FOO_BAR might have a true value of zero
587 (or in the general case an empty string or even undef)
589 2) In case FOO_BAR is not available in this platform,
590 &FOO_BAR avoids the bareword warning
592 - wait more (1.5 seconds instead of 0.1) for the CLOCK_REALTIME test
593 but expect the 'customary' slop of 0.20 instead of 0.25
594 - fixed inside a comment HAS_POLL -> TIME_HIRES_NANOSLEEP
595 - at the end of HiRest.t tell how close we were to termination
598 - Gisle noticed a mistake (using HAS_NANOSLEEP) in 1.79
601 - try nanosleep for emulating usleep -- may help in some weird
602 embedded realtime places which have nanosleep but neither usleep
603 nor select nor poll (doesn't have to be weird embedded realtime
604 place, though -- in many places usleep is nanosleep anyway)
605 - try poll for emulating usleep -- this may help some obscure/old
606 SVR4 places that have neither usleep nor select
607 - a redundant test guard in HiRes.t
610 - ITIMER_VIRTUAL detection in HiRes.t had problems (that we cannot
611 in the general case fail already at 'use' phase is suboptimal)
612 - fixes to the documentation of clock_gettime() and clock_getres()
615 - add support for the POSIX clock_gettime() and clock_getres(),
616 if available, either as library calls or as syscalls
617 - be more defensive about missing functionality: break out
618 early (during 'use') if no e.g. clock_getres() is available,
619 and protect our back by trapping those cases also in HiRes.xs
620 - the test added in 1.76 could cause an endless loop e.g. in Solaris,
621 due to mixing of sleep() and alarm() (bad programmer, no cookie!)
624 - testing for nanosleep had wrong logic which caused nanosleep
625 to become undefined for e.g. Mac OS X
626 - added a test for a core dump that was introduced by Perl 5.8.0
627 safe signals and was fixed for the time of 5.8.1 (one report of
628 the core dump was [perl #20920]), the test skipped pre-5.8.1.
629 - *cough* s/unanosleep/nanosleep/g; *cough*
632 - installation patch from Gisle Aas: in Perls 5.8.x and later
633 use MakeMaker INSTALLDIRS value of 'perl' instead of 'site'.
636 - [cpan #14608] Solaris 8 perl 5.005_03 File::Spec module does not have method rel2abs
637 (the workaround is not to use rel2abs, should not be necessary)
638 - [cpan #14642] U2time wrongly exported on the C API
639 (patch supplied by the reporter, SALVA@cpan.org)
640 - add release dates to Changes
643 - Time::HiRes::nanosleep support for Solaris [PATCH]
644 (POSIX::uname() not available if building with core perl,
645 from Gisle Aas, via perl5-porters, perl change #25295)
648 - going back to the 1.68 loader setup (using DynaLoader)
649 since too many weird things starting breaking
650 - fix a typo in José Auguste-Etienne's name
653 - a thinko in the nanosleep() detection
654 - move more changes stuff from the README to Changes
655 - add -w to the Makefile.PL
658 - oops in 1.69 about @ISA (not affecting anything but silly)
659 - add copyright 2005 to HiRes.pm
660 - add copyright and license to HiRes.xs
661 - add copyrights 2003, 2004, 2005 to README
664 - actually run a test for nanosleep
665 (if there is no $Config{d_nanosleep}) since e.g. in AIX 4.2
666 it seems that one can link in nanosleep() but then calling
667 it fails instantly and sets errno to ENOSYS (Not implemented).
668 This may be fixable in the AIX case by figuring out the right
669 (realtime POSIX?) libs and whatnot, but in the general case
670 running a real test case is better. (Of course, this change
671 will no doubt run into portability problems because of the
672 execution step...) Note that because of hysterical raisins
673 most Perls do NOT have $Config{d_nanosleep} (scanning for
674 it by Configure would in many platforms require linking in
675 things like -lrt, which would in many platforms be a bad idea
677 (from José Auguste-Etienne)
678 - support XSLoader also since it's much faster
679 (from Alexey Tourbin)
680 - add SEE ALSO (BSD::Resource and Time::TAI64)
683 - somehow 1.67 had a lot of doubled lines (a major cut-and-paste
684 error suspected), but miraculously it still worked since the
685 doubling took place below the __END__ token
686 - undef Pause() before defining it to avoid redefinition warnings
687 during compilation in case perl.h had already defined Pause()
688 (part of perl change #24271)
692 - (internal) don't ignore the return value of gettimeofday()
693 - (external) return undef or an empty if the C gettimeofday() fails
694 (affects Time::HiRes gettimeofday() and the hires time())
698 - fix the 'hierachy' typo in Makefile.PL [rt.cpan.org #8492]
699 - should now build in Solaris [rt.cpan.org #7165] (since 1.64)
700 - should now build in Cygwin [rt.cpan.org #7535] (since 1.64)
701 - close also [rt.cpan.org #5933] "Time::HiRes::time does not
702 pick up time adjustments like ntp" since ever reproducing it
703 (and therefore verifying a possible fix) in the same environment
704 has become rather unlikely
707 - one should not mix u?alarm and sleep (the tests modified
708 by 1.65, #12 and #13, hung in Solaris), now we just busy
709 loop executing an empty block
710 - in the documentation underline the unspecificity of mixing
712 - small spelling fixes
715 - regenerate ppport.h with Devel::PPPort 3.03,
716 now the MY_CXT_CLONE is defined in ppport.h,
717 we no more need to do that.
719 - the test #12 would often hang in sigsuspend() (at least that's
720 where Mac OS X' ktrace shows it hanging). With the sleep()s
721 changed to sleep(1)s, the tests still pass but no hang after
722 a few hundred repeats.
725 - Win32 and any ithread build: ppport.h didn't define
726 MY_CXT_CLONE, which seems to be a Time-HiRes-ism.
729 - Skip testing if under PERL_CORE and Time::HiRes has not
730 been Configured (from Marcus Holland-Moritz, core change
732 - Use ppport.h generated by Devel::PPPort 3.01,
733 allowing cutting away our own portability code.
734 - Don't use $ENV{PERL_CORE} for < 5.6.0.
735 - Don't use "for my $i" for <= 5.003.
736 - Don't use Pause() for <= 5.003.
737 - Can't use newSVpvf for <= 5.003.
738 (most of the changes from Marcus)
741 - Win32: reset reading from the performance counters every
742 five minutes to better track wall clock time (thanks to
743 PC timers being often quite bad), should help long-running
747 - Win32: Patch from Steve Hay
748 [PATCH] Re: [perl #30755] [Win32] Different results from Time::HiRes::gettimeofdayunder the debugger
749 to [perl #30755] reported by Nigel Sandever
751 - Cygwin: Use the Win32 recalibration code also in Cygwin if the
752 <w32api/windows.h> APIs are available. Cygwin testing by
753 Yitzchak Scott-Thoennes.
755 - Solaris: use -lposix4 to get nanosleep for Solaris 2.6,
756 after that keep using -lrt, patch from Alan Burlison,
757 bug reported in [cpan #7165]
760 - Change the Win32 recalibration limit to 0.5 seconds and tweak
761 the documentation to blather less about the gory details of the
762 Win32 implementation and more about the complications in general
763 of meddling with the system clock.
766 - Document the 1.57 change better.
769 - Win32/Cygwin/MinGW: if the performance counter drifts by more
770 than two seconds from the system clock (due to ntp adjustments,
771 for example), recalibrate our internal counter: from Jan Dubois,
772 based on [cpan #5933] by Jerry D. Hedden.
775 - Give a clearer message if the tests timeout (perl change #22253)
776 - Don't use /tmp or its moral equivalents (perl bug #15036,
780 - Windows: mingw32 patch from Mike Pomraning (use Perl's Const64()
781 instead of VC-specific i64 suffix)
784 - Solaris: like Tru64 (dec_osf) also Solaris need -lrt for nanosleep
787 - Windows: higher resolution time() by using the Windows
788 performance counter API, from Jan Dubois and Anton Shcherbinin.
789 The exact new higher resolution depends on the hardware,
790 but it should be quite a bit better than using the basic
794 - In AIX (v?) with perl 5.6.1 the HiRes.t can hang after
795 the subtest 18. No known analysis nor fix, but added
796 an alarm (that requires fork() and alarm()) to the test.
799 - doc tweaks from mjd (perl change #20456)
800 - NCR MP-RAS hints file added (svr4.pl) (perl change #21249)
803 - add a message (for non-core builds) to Makefile.PL about
804 the LC_ALL=C workaround
805 - &Time::HiRes::d_nanosleep was broken (perl change #20131)
806 - the nanosleep() probe was broken (perl change #20061)
807 - use existence instead of definedness for feature probes
809 - MPE/iX tweak (perl change #20042)
810 - do not use HAS_NANOSLEEP (perl change #19898)
813 - UVuf for non-IVSIZE platforms (from Keiichiro Nagano)
814 - OS/2 can always mix subsecond sleeps with signals
815 (part of perl change #19789)
818 - workaround for buggy gcc 2.95.3 in openbsd/sparc64
822 - do not use -lrt in Linux (from March Lehmann, perl change #19449)
823 - unnecessary (nanosleep is in libc anyway)
824 - harmful (-lrt slows down execution)
825 - incompatible (with many distributions' pthreads)
828 - do not create files in blib directories under core
829 (perl change #19160, from rgs)
830 - detypo s/VTLARM/VTARLM/ (perl change #19328, from mjd)
833 - guarantee that $xdefine in HiRes.t is always defined
834 (perl change #19109, from IlyaZ)
835 - a cleaner way to detect PERL_CORE (perl change #19111,
839 - add hints/irix.pl to turn off overly POSIX flags that
840 cause hide struct timespec to be hidden (and compilation
841 to fail) (bleadperl change #19085)
842 - documentation tweaks
845 - add c:/temp to the list of temp directories to probe
846 so that cygwin (and win*?) builds are happy. This was
847 needed at least in my cygwin 1.3.20/w2k setup.
850 - modernize the constants code (from Nicholas Clark)
853 - At some point the ability to figure our the correct incdir
854 for EXTERN.h (either a core perl build, or an installed perl)
855 had broken (which lead into all test compiles failing with
856 a core perl build, but thanks to the robustness of Makefile.PL
857 nothing of this was visible). The brokenness seemed to be
858 caused by $ENV{PERL_CORE} not being on for core builds?
859 Now stole a trick from the Encode that sets $ENV{PERL_CORE}
860 right, and both styles of build should work again.
863 - Nicholas Clark noticed that the my_catdir() emulation function
864 was broken (which means that we didn't really work for Perls
866 - inspired by fixing the above made the whole Makefile.PL -w
868 - tightened up the Makefile.PL output, less whitespace
871 - fix from Craig Berry for better building in VMS with PERL_CORE
874 - no functional changes
875 - move lib/Time/HiRes.pm as Hires.pm
876 - libraries scanning was slightly broken (always scanned
877 for a library even when $Config{libs} already had it)
880 - Ray Zimmerman ran into a race condition in Mac OS X.
881 A 0.01-second alarm fired before the test expected.
882 The test first slept indefinitely (blocking for signals)
883 and only after that tested for the signal having been sent.
884 Since the signal had already been sent, the test #12 never
885 completed. The solution: test first, then block.
886 - default to being silent on all probing attempts, set the
887 environment variable VERBOSE to a true value to see the
888 details (the probing command and the possible errors)
891 - do not clear MAN3PODS in Makefile.PL (Radoslaw Zielinski)
892 - INSTALLDIRS => 'perl' missing which means that Time::HiRes
893 cannot be upgraded from CPAN to override the 5.8.0 version
895 - Time::HiRes 1.35 could not be dropped as-is to bleadperl
896 because the include directories did not adjust themselves
897 if $ENV{PERL_CORE} (Hugo van der Sanden)
898 - add documentation about the restart of select() under alarm()
901 - small documentation tweaks
905 - better VMS operation (Craig Berry)
908 - our time machine is accelerating: now works with Perl 5.004_01
909 (tried with 5.003_07 and 5.002 but I get segmentation faults
910 from running the Makefile.PL with those in Tru64 4.0D)
913 - backward compatibility (pre-5.6.0) tweaks:
914 - no XSLoader in 5.00503, use DynaLoader instead
915 - no SvPV_nolen, either
916 - no PerlProc_pause(), either
917 - now tested with 5.00404 and 5.00503
918 - Makefile.PL requires 5.00404 (no more 5.002)
919 - use nanosleep instead of usleep, if it is available (Wilson Snyder)
920 (this means that one can mix subsecond sleeps with alarms)
921 - because of nanosleep we probe for -lrt and -lposix4
922 - the existence of getitimer/nanosleep/setitimer/ualarm/usleep
923 is available by exportable constants Time::HiRes::d_func
924 (since older Perl do not have them in %Config, and even
925 5.8.0 does not probe for nanosleep)
928 - backward compatibility (pre-5.6.1) tweaks:
929 - define NV if no NVTYPE
930 - define IVdf if needed (note: the Devel::PPPort
931 in 5.8.0 does not try hard hard enough since
932 the IVSIZE might not be defined)
933 - define NVgf if needed
934 - grab the typemap from 5.8.0 for the NV stuff
936 1.31 and 1.32 add more backward compatibility (now all the way
937 back to Perl 5.00404), and using nanosleep() (if available) for
942 - release 1.29_02 as 1.30
944 1.30 adds all the changes made during the Perl 5.6->5.7->5.8
945 development cycle. Most notably portability across platforms has been
946 enhanced, and the interval timers (setitimer, getitimer) have been
947 added. Note that the version of Time::HiRes that is included in Perl
948 5.8.0 calls itself 1.20_00, but it is equivalent to this Time::HiRes
949 version. Note also that in 1.30 Wegscheid turns over the maintenance
950 to Jarkko Hietaniemi.
954 - fix a silly unclosed comment typo in HiRes.xs
955 - document and export REALTIME_REALPROF (Solaris)
959 - only getitimer(ITIMER_REAL) available in Cygwin and Win32
960 (need to patch this also in Perl 5.[89])
961 - remove CVS revision log from HiRes.xs
965 The following numbered patches refer to the Perl 5.7 changes,
966 you can browse them at http://public.activestate.com/cgi-bin/perlbrowse
968 - 17558: Add #!./perl to the .t
969 - 17201: linux + usemorebits fix, from Rafael Garcia-Suarez
970 - 16198: political correctness, from Simon Cozens
971 - 15857: doc tweaks, from Jarkko Hietaniemi
972 - 15593: optimization in .xs, from Paul Green
973 - 14892: pod fixes, from Robin Barker
974 - 14100: VOS fixes, from Paul Green
975 - 13422: XS segfault, from Marc Lehmann
976 - 13378: whether select() gets restarted on signals, depends
977 - 13354: timing constraints, again, from Andy Dougherty
978 - 13278: can't do subsecond alarms with ualarm;
979 break out early if alarms do not seem to be working
980 - 13266: test relaxation (cygwin gets lower hires
981 times than lores ones)
982 - 12846: protect against high load, from Jarkko Hietaniemi
983 - 12837: HiRes.t VMS tweak, from Craig A. Berry
984 - 12797: HiRes.t VMS tweak, from Charles Lane
985 - 12769: HiRes.t VMS tweak, from Craig A. Berry
986 - 12744: gcc vs MS 64-bit constant syntax, from Nick Ing-Simmons
987 - 12722: VMS ualarm for VMS without ualarm, from Charles Lane
988 - 12692: alarm() ain't gonna work if ualarm() ain't,
989 from Gurusamy Sarathy
990 - 12680: minor VMS tweak, from Charles Lane
991 - 12617: don't try to print ints as IVs, from Jarkko Hietaniemi
992 - 12609: croak on negative time, from Jarkko Hietaniemi
993 - 12595: Cygwin rounds up for time(), from Jarkko Hietaniemi
994 - 12594: MacOS Classic timeofday, from Chris Nandor
995 - 12473: allow for more than one second for sleep() and usleep()
996 - 12458: test tuning, relax timing constraints,
997 from Jarkko Hietaniemi
998 - 12449: make sleep() and usleep() to return the number
999 of seconds and microseconds actually slept (analogously
1000 with the builtin sleep()), also make usleep() croak if
1001 asked for more than 1_000_000 useconds, from Jarkko Hietaniemi
1002 - 12366: Time::HiRes for VMS pre-7.0, from Charles Lane
1003 - 12199: do not use ftime on Win32, from Gurusamy Sarathy
1004 - 12196: use ftime() on Win32, from Artur Bergman
1005 - 12184: fix Time::HiRes gettimeofday() on Win32, from Gurusamy Sarathy
1006 - 12105: use GetSystemTime() on Win32, from Artur Bergman
1007 - 12060: explain the 1e9 seconds problem, from Jarkko Hietaniemi
1008 - 11901: UNICOS sloppy division, from Jarkko Hietaniemi
1009 - 11797: problem in HiRes.t, from John P. Linderman
1010 - 11414: prototype from Time::HiRes::sleep(), from Abhijit Menon-Sen
1011 - 11409: Time::HiRes qw(sleep) failed, from Abhijit Menon-Sen
1012 - 11270: dynix/ptx 4.5.2 hints fix, from Peter Prymmer
1013 - 11032: VAX VMS s/div/lib\$ediv/ fix, from Peter Prymmer
1014 - 11011: VAX VMS s/qdiv/div/ fix, from Peter Prymmer
1015 - 10953: SCO OpenServer 5.0.5 requires an explicit -lc for usleep(),
1017 - 10942: MPE/IX test tweaks, from Mark Bixby
1018 - 10784: unnecessary pod2man calls, from Andy Dougherty
1019 - 10354: ext/ + -Wall, from Doug MacEachern
1020 - 10320: fix the BOOT section to call myU2time correctly
1021 - 10317: correct casting for AIX< from H. Merijn Brand
1022 - 10119: document that the core time() may be rounding, not truncating
1023 - 10118: test fix, from John Peacock
1024 - 9988: long =item, from Robin Barker
1025 - 9714: correct test output
1026 - 9708: test also the scalar aspect of getitimer()
1027 - 9705: Add interval timers (setitimer, getitimer)
1028 - 9692: do not require at least 5.005 using XS
1030 The following changes were made on top of the changes
1031 made for Time::HiRes during the Perl 5.7 development
1032 cycle that culminated in the release of Perl 5.8.0.
1034 - add "require 5.005" to the Makefile.PL
1035 - remove the REVISION section (CVS log) from HiRes.pm
1036 - add jhi's copyright alongside Douglas'
1037 - move HiRes.pm to lib/Time/
1038 - move HiRes.t to t/
1039 - modify HiRes.t to use $ENV{PERL_CORE}
1040 - modify the original Time::HiRes version 1.20 Makefile.PL
1041 to work both with Perl 5.8.0 and the new code with pre-5.8.0
1042 Perls (tried with 5.6.1)
1043 - tiny tweaks and updates in README and TODO
1044 - bump the VERSION to 1.29
1046 1.20 Wed Feb 24 21:30 1999
1047 - make our usleep and ualarm substitutes into hrt_usleep
1048 and hrt_ualarm. This helps static links of Perl with other
1049 packages that also have usleep, etc. From
1050 Ilya Zakharevich <ilya@math.ohio-state.edu>
1051 - add C API stuff. From Joshua Pritikin
1052 <joshua.pritikin@db.com>
1053 - VMS Makefile.PL fun. From pvhp@forte.com (Peter Prymmer)
1054 - hopefully correct "-lc" fix for SCO.
1057 1.20 adds a platform neutral set of C accessible routines if you are
1058 running 5.005+. All other changes are packaging changes and build
1059 fixes(?) for statically linked Perl, SCO, and VMS.
1061 1.19 Tue Sep 29 22:30 1998
1062 - put VMS gettimeofday() in. Patch is from Sebastian Bazley
1063 <seb@stian.demon.co.uk>
1064 - change GIMME_V to GIMME to help people with older versions of
1066 - fix Win32 version of gettimeofday(). It didn't affect anything,
1067 but it confuses people reading the code when the return value
1068 is backwards (0 is success).
1069 - fix Makefile.PL (more) so that detection of gettimeofday is
1072 1.19 has better VMS support.
1074 1.18 Mon Jul 6 22:40 1998
1075 - add usleep() for Win32.
1076 - fix Makefile.PL to fix reported HP/UX feature where unresolved
1077 externals still cause an executable to be generated (though no
1078 x bit set). Thanks to David Kozinn for report and explanation.
1079 Problems with the fix are mine :)
1081 1.18 has limited Win32 support (no ualarm). Added usleep for Win32.
1082 Probably buggy. I'm sure I'll hear.
1084 1.17 Wed Jul 1 20:10 1998
1085 - fix setitimer calls so microseconds is not more than 1000000.
1086 Hp/UX 9 doesn't like that. Provided by Roland B Robert, PhD.
1087 - make Win32. We only get gettimeofday (the select hack doesn't
1088 seem to work on my Win95 system).
1089 - fix test 4 on 01test.t. add test to see if time() and
1090 Time::HiRes::time() are close.
1092 1.16 Wed Nov 12 21:05 1997
1093 - add missing EXTEND in new gettimeofday scalar code.
1095 1.16+ should be closer to building out of the box on Linux. Thanks
1096 to Gisle Aas for patches, and the ualarm equivalent using setitimer.
1098 If your underlying operating system doesn't implement ualarm(), then
1099 a fake using setitimer() will be made. If the OS is missing usleep(),
1100 a fake one using select() will be made. If a fake can't be made for
1101 either ualarm() or usleep(), then the corresponding Perl function will
1102 not be available. If the OS is missing gettimeofday(), you will get
1103 unresolved externals, either at link- or run-time.
1105 This is an improvement; the package used to not even build if
1106 you were missing any of these bits. Roderick Schertler
1108 <roderick@argon.org> did all the conditional compilation stuff,
1109 look at HiRes.pm and the test suites; it's good educational reading.
1111 1.15 Mon Nov 10 21:30 1997
1112 - HiRes.pm: update pod. Provided by Gisle Aas.
1113 - HiRes.xs: if gettimeofday() called in scalar context, do
1114 something more useful than before. Provided by Gisle Aas.
1115 - README: tell of xsubpp '-nolinenumber' woes. thanks to
1116 Edward Henigin <ed@texas.net> for pointing out the problem.
1118 1.14 Wed Nov 5 9:40 1997
1119 - Makefile.PL: look for setitimer
1120 - HiRes.xs: if missing ualarm, but we have setitimer, make up
1121 our own setitimer. These were provided by Gisle Aas.
1123 1.13 Tue Nov 4 23:30 1997
1124 - Makefile.PL: fix autodetect mechanism to do try linking in addition
1125 to just compiling; should fix Linux build problem. Fix was provided
1128 1.12 Sun Oct 12 12:00:00 1997
1129 - Makefile.PL: set XSOPT to '-nolinenumbers' to work around xsubpp bug;
1130 you may need to comment this back out if you have an older xsubpp.
1131 - HiRes.xs: set PROTOTYPES: DISABLE
1133 1.11 Fri Sep 05 16:00:00 1997
1135 Had some line commented out that shouldn't have been (testing
1138 Previous version was corrupted.
1140 1.10 Thu May 22 20:20:00 1997
1141 - HiRes.xs, HiRes.pm, t/*:
1142 - only compile what we have OS support for (or can
1144 - only test what we compiled
1145 - gross improvement to the test suite
1147 This work was all done by Roderick Schertler
1148 <roderick@argon.org>. If you run Linux or
1149 one of the other ualarm-less platforms, and you like this
1150 module, let Roderick know; without him, it still wouldn't
1151 be working on those boxes...
1152 - Makefile.PL: figure out what routines the OS has and
1153 only build what we need. These bits were written by Jarkko
1154 Hietaniemi <jhi@iki.fi>. Again, gratitude is due...
1156 1.02 Mon Dec 30 08:00:00 1996
1157 - HiRes.pm: update documentation to say what to do when missing
1158 ualarm() and friends.
1159 - README: update to warn that ualarm() and friends need to exist
1161 1.01 Fri Oct 17 08:00:00 1996
1162 - Makefile.PL: make XSPROTOARGS => '-noprototyopes'
1163 - HiRes.pm: put blank line between __END__ and =head1 so that
1166 1.00 Tue Sep 03 13:00:00 1996
1167 - original version; created by h2xs 1.16